Object Image Size Is a Fundamental Coding Dimension in Human Vision: New Insights and Model

Area of Science:

• Visual perception
• Psychophysics
• Computational neuroscience

Background:

• Luminance contrast is integrated over retinal area via contrast gain control.
• This integration process, if varied across retinal regions, could form the basis for visual size coding.

Purpose of the Study:

• To investigate if distinct visual mechanisms contribute to size coding.
• To test a novel model of visual size perception accounting for contrast gain control and multiple size mechanisms.

Main Methods:

• Employed a standard adaptation paradigm with novel features: spatially jittering adaptors and decoupled adaptor/target spatial frequencies.
• Used psychophysical experiments with human participants (N=8) to measure size aftereffects.

Main Results:

• Observed a bidirectional size adaptation aftereffect, indicating robust size coding.
• Found that standard population coding models are insufficient; developed a new model with multiple size mechanisms and divisive surround-suppression.
• Identified asymmetries in size aftereffects, with greater effects for smaller targets and incomplete return to baseline.

Conclusions:

• The visual system employs a novel coding strategy for size, distinct from contrast coding.
• The proposed model, incorporating multiple size mechanisms and surround-suppression, accurately explains the observed psychophysical results.
• These findings advance our understanding of visual coding dimensions and strategies.